Point cloud

What is a point cloud?

A point cloud is a large set of individual measured points in three-dimensional space. Each point records one spot where a sensor detected a surface, and it carries an X, Y and Z coordinate that places it in a shared coordinate system. Usually each point also stores a colour, and often an intensity value that describes how strongly the surface returned the sensor's signal. A point cloud has no faces, edges or surfaces of its own. It is raw measurement, and the shape of the building only becomes visible when enough points are viewed together.

In Slovak practice the term is written as mračno bodov. When the coordinates are tied to a real-world reference system, the data is often called a georeferenced point cloud. It arrives as a file that a viewer or modelling program can read, and it is the starting point for almost every later step of a survey.

How is a point cloud captured?

Point clouds come from two main families of technique. Laser scanning fires laser pulses and measures how long each one takes to return, which gives the distance to every surface it hits. Terrestrial scanners stand on a tripod, while mobile mapping systems and drones carry the sensor along a route. Photogrammetry works from photographs instead. Software finds the same features across overlapping images and calculates where they sit in space. The two methods are not interchangeable, and the table shows where each one tends to fit.

What can you do with a point cloud?

The most common first use is viewing. Once the cloud is open, you can walk through a room, check whether a wall leans, or see where a ceiling sags, without returning to the site. Architects then cut sections through the cloud at chosen heights and positions. This gives a true cross-section of the building rather than one drawn from memory. Plans, elevations and details are traced over those sections, either by hand in CAD or by drawing directly on the cloud in a modelling program.

Modelling is the step that turns the cloud into something a design team can edit. Walls, floors, openings and roof planes are fitted to the points, and the result is a BIM model or a CAD drawing. This fitting is manual or semi-automatic work, and the quality of the final model depends heavily on the care of the person doing it.

Which file formats carry a point cloud?

Point clouds are exchanged in a small group of formats, and the choice matters when a survey firm delivers data to a design office that uses different software. The table summarises the ones you are most likely to meet.

E57 is the most portable choice in principle, but readers differ in how strictly they follow the standard. A file that opens in one program may need testing in another before the project relies on it. Name the format in the survey brief, and ask for a test file before the engagement starts.

How does a point cloud differ from a BIM model?

A point cloud records what is physically there, with every point measured and none of them labelled. A BIM model records building elements such as walls, slabs and doors, each with a type, a material and relationships to its neighbours. The cloud is evidence and the model is an interpretation of that evidence. A cloud cannot tell you that a surface is a load-bearing wall or that a window has a particular frame. A model can, but only if someone has put that knowledge into it.

This is why the two are usually produced in sequence. The cloud comes first, and the modelling step is what the industry calls scan-to-BIM. Keep the cloud file in the project archive even after the model is finished, because it is the only record of the building as measured and it allows checks against the model later.

How does a point cloud differ from an as-built survey?

An as-built survey is a formal record of the completed building against its plan, prepared by a licensed surveyor for permitting or the cadastre. A point cloud can be part of the evidence behind such a survey, but it is not the survey itself. The authority does not accept the cloud as the document, so the official drawings must still be produced and signed by the surveyor. Confusing the two is a common mistake when a homeowner receives a cloud and assumes the paperwork is finished.

Why does raw point cloud data need cleaning before use?

Raw clouds contain things nobody wanted to measure. Moving people, reflections off glass and water, scaffolding and dust all leave stray points, and they can make a clean wall look thick or a floor look uneven. Cleaning removes that noise, and it is normally done before any section or model is drawn from the data.

Registration is the other essential step. Each scan position produces its own cloud, and these must be aligned into one coordinate system using overlaps or reference targets. Poor registration shows up as doubled walls and misaligned floors, so check the joins at several places in the building rather than trusting the software's summary.

What misconceptions should a homeowner know about point clouds?

The first misconception is that a point cloud shows what lies behind the walls. It records only surfaces the sensor could see, so pipes, cables and voids inside a wall stay invisible unless they were exposed during the survey. The second is that colour in a cloud is a true photograph. Colour is attached point by point from a camera, so it depends on lighting and can be uneven across a single room.

The third misconception is that more points always mean more accuracy. A dense cloud from badly planned scan positions can be less useful than a sparser cloud captured with care. Coverage, target placement and registration usually matter more than the raw density of the file.

• A scan taken from one position leaves shadows behind furniture and in corners.
• Glass, mirrors and water can return misleading points.
• Colour can change between two areas of the same wall.